The diffusivity of phosphorus in San Carlos olivine (SCO) was measured at near-atmospheric pressure and 650–850 °C by in-diffusion of P from a surface powder source consisting of pre-reacted SCO and AlPO 4 . The experiments were conducted in evacuated silica-glass ampoules at oxygen fugacities fixed by solid-state buffers, generally Ni-NiO but also including two experiments buffered at wustite-magnetite. Phosphorus uptake profiles were characterized by Rutherford backscattering spectroscopy (RBS) and nuclear reaction analysis (NRA). The temperature dependence of P diffusion in SCO conforms to the expected Arrhenius relation D = D 0 exp(− E a /R T ), where the constants are as follows: log( D 0 , m 2 /s) = −10.06 ± 0.80 and E a = 229 ± 16 kJ/mol. These values characterize P as a relatively slow diffuser in olivine—slower by about an order of magnitude than Cr and Ca at basalt near-liquidus temperatures—but substantially faster than Si. With a view toward modeling P uptake during rapid growth of natural olivines, P diffusion was also characterized in dry MORB basalt melt over the temperature range 1250–1500 °C at 1 GPa, using traditional diffusion couples contained in graphite. Phosphorus diffusion profiles in the quenched and depressurized samples were quantified by laser-ablation ICP/MS. Phosphorus diffusion in basaltic melt is similar to that of Si, with log( D 0 , m 2 /s) = −6.30 ± 0.7 and E a = 147 ± 22 kJ/mol. The new data for P diffusion in olivine and basalt melt can be used to explore the acquisition of fine-scale zoning in natural olivine phenocrysts through kinetic models, as well as the survival of P zoning in olivine with time spent at elevated temperature. Models of growth entrapment of a P-enriched near-surface layer in the olivine lattice indicate that crystal growth at plausible sustained rates is indeed likely to result in regions of anomalously high P content in the resulting crystal. Phosphorus concentrations above the equilibrium partitioning value can also result from development of a diffusive boundary layer in the melt against a rapidly growing crystal, but this mechanism is ineffective at typical sustained olivine growth rates, requiring dendrite-forming growth speeds. Preservation of P zoning on the scale of a few micrometers apparently requires cooling within a few months of formation of the zoning.